G-quadruplex Stabilization Fuels the ALT Pathway in ALT-positive Osteosarcoma Cells

Most human tumors maintain telomere lengths by telomerase, whereas a portion of them (10–15%) uses a mechanism named alternative lengthening of telomeres (ALT). The telomeric G-quadruplex (G4) ligand RHPS4 is known for its potent antiproliferative effect, as shown in telomerase-positive cancer models. Moreover, RHPS4 is also able to reduce cell proliferation in ALT cells, although the influence of G4 stabilization on the ALT mechanism has so far been poorly investigated. Here we show that sensitivity to RHPS4 is comparable in ALT-positive (U2OS; SAOS-2) and telomerase-positive (HOS) osteosarcoma cell lines, unlinking the telomere maintenance mechanism and RHPS4 responsiveness. To investigate the impact of G4 stabilization on ALT, the cardinal ALT hallmarks were analyzed. A significant induction of telomeric doublets, telomeric clusterized DNA damage, ALT-associated Promyelocytic Leukaemia-bodies (APBs), telomere sister chromatid exchanges (T-SCE) and c-circles was found exclusively in RHPS4-treated ALT cells. We surmise that RHPS4 affects ALT mechanisms through the induction of replicative stress that in turn is converted in DNA damage at telomeres, fueling recombination. In conclusion, our work indicates that RHPS4-induced telomeric DNA damage promotes overactivation of telomeric recombination in ALT cells, opening new questions on the therapeutic employment of G4 ligands in the treatment of ALT positive tumors.


Introduction
Telomeres are nucleoprotein structures that protect the ends of linear chromosomes [1,2]. In humans, telomeres are composed of 10-15 kb of tandem arrays of TTAGGG repeats [3,4] bound by specialized proteins [5] that cooperatively conceal chromosome ends, impeding their processing as DNA double-strand breaks (DSBs) and contributing to genome stability. In cells lacking a telomere maintenance mechanism (TMM), telomeric DNA encounters progressive erosion at each cell duplication, eventually leading to replicative senescence. In somatic primary cells, this represents a tumor-suppressive mechanism and, as a consequence, continuous dividing cells (e.g., cancer cells) need to avoid senescence by the maintenance of stable telomere lengths. Most human cancers maintain their telomeres by reactivating the reverse transcriptase telomerase [6]. However, a significant fraction activity (TA) was tested in all the cell lines by TRAP-assay, confirming that only HOS cells are telomerase-positive. All the aforementioned cell lines were maintained at 37 • C in a 5% CO2 and 95% air atmosphere. All cell lines were routinely mycoplasma tested.
If not differently stated, all the experiments were performed treating cells with their respective IC 50 calculated 120 h after treatment.

Growth Curves and cPDL Calculation
Cells were seeded at the density of 5000 cells/cm 2 and treated using increasing concentrations of RHPS4. After 72, 96, and 120 h cells were detached and counted. Cumulative population doubling level (cPDL) was calculated as follows: cPDL = log 2(N f /N 0 ), where N f is the final cell number and N 0 is the initial number of seeded cells. Results were obtained from three independent experiments.

Sulforhodamine B (SRB) Assay
Exponential growing cells were harvested, counted, and seeded in 96-well plates at different density for each line (HOS: 500 cells/well; U2OS: 1000 cells/well; SAOS: 4000 cells/well). Optimal seeding density was determined to ensure exponential growth during a 5-day assay (120 h). The SRB assay was performed as previously described [44], with minor modifications. Cell were fixed in 10% cold trichloroacetic acid, incubated at 4 • C for 1 h and then washed with deionized water. Cells were stained with 200 µL/well of 0.1% SRB (Sigma-Aldrich, Co., St. Louis, MO, USA) for 30 min and washed four times with 1% acetic acid. Plates were air-dried at room temperature (RT), and stained proteins were solubilized with 200 µL/well of 10 mM unbuffered Tris base (tris(hydroxymethyl) aminomethane) (Sigma-Aldrich Co., St. Louis, MO, USA). Optical density was read at 530 nm with a Victor plate-reader (VICTOR X3 Multilabel plate reader, PerkinElmer, Waltham, MA, USA). Experiments were repeated five times.

Flow Cytometric Analysis
Cells were treated with RHPS4 for 72, 96 and 120 h, pulsed with 10 µM bromodeoxyuridine (BrdU) in the last 30 min and then fixed and analyzed. Briefly, each sample was fixed, permeabilized, and the histones were dissociated with 2 M HCl as previously described [30]. BrdU positive cells were detected with an anti-BrdU primary antibody diluted 1:100 (DAKO Cytomatation, Glostrup, Denmark) and with an anti-mouse-Alexa488 conjugated diluted 1:100 (Invitrogen, Life Technologies, Carlsbad, CA, USA). Both antibodies were incubated for 1 h RT in the dark. All samples were counterstained with propidium iodide (PI) for DNA/BrdU biparametric analysis. BrdU positive cells were gated, and the relative percentage was calculated by Cytexpert software (Beckman Coulter, Inc., San Diego, CA, USA).

Collection of Chromosome Spreads
Chromosome spreads were obtained following 4 h incubation in 5 × 10 -6 M colchicine (Sigma-Aldrich). Spreads of chromosomes were prepared following standard procedure consisting of treatment with a hypotonic solution (75 mM KCl) for 20 min at 37 • C, followed by fixation in freshly prepared Carnoy solution (3:1 v/v methanol/acetic acid). Cells were then dropped onto slides, air-dried and utilized for cytogenetic analysis.

Chromosome Orientation-FISH (CO-FISH) Analysis
Cell lines subcultured in the presence of 5'-bromo-2'-deoxyuridine (BrdU, Sigma Aldrich) at a final concentration of 2.5 × 10 -5 M and were then allowed to replicate their DNA once at 37 • C overnight (ON). Cells were then collected, and chromosome spreads were prepared as described above. CO-FISH was performed as previously described [45] using a (TTAGGG) 3 probe labeled with Cy3 and a (CCCTAA) 3 probe labeled with FITC (Panagene, Yuseong-gu, Korea). Images were captured with an Axio Imager.M1 equipped with a CCD camera. T-SCEs were scored only when the double signals were visible with both the Cy3 and FITC probes. Experiments were repeated three times and 4000 chromosome ends were analyzed for each line and condition. G-SCE was evaluated by scoring the number of chromosomes with normal (trans) and recombined (cis) CO-FISH signals configuration. To correct cis frequencies for multiple crossovers, we used the following expression [46]: which is equivalent to in which the frequency of SCEs that would be necessary to account for the number of cis events represents the mean (µ) of a Poisson distribution, whose sum of odd-numbered terms equals the observed cis frequency (γ). The latter equation was used to convert the observed cis frequencies into estimates of the true crossover frequency, also referred to as the SCE equivalent frequency. For this purpose, 4000 chromosome ends were scored in three independent experiments. Moreover, double color telomere doublets, a marker of telomere fragility, were counted for each chromosome end considering both the following configuration: overlapping double green and red signals; red signal beside a green signal.

TIF Co-immunostaining
Cells were fixed with 4% paraformaldehyde (PFA) at 4 • C, permeabilized with 0.2% Triton X-100 at 4 • C and blocked in BSA 1% dissolved in phosphate-buffered saline (PBS) (w/v) at 37 • C. Samples were then co-immunostained overnight (ON) at 4 • C, using a rabbit polyclonal telomeric protein TRF1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) in combination with a mouse monoclonal anti-53BP1 antibody (Millipore Corp. 290 Concord Rd, Billerica, MA, USA). After washes in PBS/BSA 1% samples were incubated with the secondary antibody (anti-mouse Alexa 546 and anti-rabbit Alexa 488, Invitrogen, Life Technologies, Carlsbad, CA, USA) for 1 h at 37 • C. Finally, slides were washed in PBS/BSA 1%, counterstained with DAPI (Sigma-Aldrich), and images were acquired with an Axio-Imager.M1 fluorescent microscope (Carl Zeiss, Jena, Germany) and analyzed using ISIS software (Metasystems, Milano, Italy). The frequency of DNA damage marker foci and TRF1/53BP1 colocalization dots per cell were scored in 100 nuclei in three independent experiments.

Immuno-FISH Experiments
Immunostaining with a rabbit polyclonal antibody against PML (E-11:sc-377390, Santa Cruz Biotechnology) or RPA32/RPA2 (phosphor-S33) (ab211877, Abcam, Cambridge, MA, USA) was performed as described above. After washing with 0.05% Triton X-100 in PBS for 5 min, cells were incubated with a secondary (goat) anti-rabbit antibody conjugated with Alexa 488 diluted in blocking solution for 1 h at room temperature (for PML or RPA2 detection). After immunostaining, telomeric FISH was performed as described in the previous paragraph. Images were taken using an Axio Imager.M1 (Carl Zeiss, Jena, Germany) equipped with a CCD camera and then analyzed using ISIS software (Metasystems, Milano, Italy). At least 100 nuclei in three independent experiments were analyzed to identify colocalization events.

Real Time qPCR
Real-time qPCR analysis of TERRA was performed as described [49]. Briefly, RNA was extracted from cells with RNAeasy mini kit (Qiagen, Hilden, Germany) and accurately digested with the RNase-free DNase set (Qiagen). Then, RNA quality was checked on FA gel electrophoresis and amplified in real-time PCR assay with subtelomere-specific primers with a 7900 HT Fast Real-Time PCR System (Applied Biosystem, Waltham, MA, USA).

C-circle Assay
Total DNAs were extracted using Quick C-Circle Lysis Buffer (50 mM KCl, 10 mM Tris HCl pH 8.5, 2 mM MgCl2, 0.5% NP40, 0.5% Tween) and treated with 0.5 µg/µL protease (Qiagen). Samples were placed in thermomixer at 1400 rpm at 56 • C for 1 h than 70 • C for 20 min. Samples, 100 ng in 10-µL each, were combined with 10 µL 0.2 mg/ml BSA, 10% Tween, 100 mM each dATP, dGTP, and dTTP, 1× Φ29 buffer and with or without 7.5 U Φ29 DNA polymerase (NEB) and incubated at 30 • C for 4 h then 70 • C for 20 min. For quantification, the reaction products were dot-blotted onto a 2× SSC-soaked nylon positively charged membrane (GE Healthcare UK Limited, Little Chalfont, UK). DNA was UV-cross-linked onto the membrane, which was then hybridized with a 32P-labeled probe [T3AG2] in Church buffer (0.5 M phosphate buffer pH 7.2, 7% SDS, 1 mM EDTA, 0.1% BSA) overnight at 55 • C. The gel was washed twice and exposed to a PhosphorImager screen and analyzed by Quantity One software (Biorad, Hercules, CA, USA). Signals of samples without Φ29 DNA polymerase were subtracted from signal obtained from corresponding samples with Φ29 DNA polymerase to determine the c-circle expression value.

RHPS4 Exerts Cytotoxic Effect Over Osteosarcoma Cell Lines Independently From TMM
RHPS4 treatment (concentrations ranging from 0.5 to 2 µM) induced a significant and concentration-dependent inhibition of cell growth in both ALT and telomerase positive osteosarcoma cell lines. Growth curves indicated overlapping cell growth inhibition in U2OS, SAOS-2, and HOS independently from their intrinsic doubling times (DT) (DT lower in HOS and U2OS than in SAOS-2; Figure 1a). As shown by IC 50 values (concentrations of compound leading to 50% inhibition of cell proliferation), calculated from the dose-response curves obtained at 120 h after treatment (Figure 1b), sensitivity to the G4 ligand was comparable in ALT and telomerase positive cell lines (IC 50 values: 1.6, 1.4, 1.2 µM for SAOS-2, U2OS, and HOS, respectively).
Progression through cell cycle, and in particular S-phase, was not affected in any of the cell lines analyzed by the treatment with the IC 50 of the compound. This data suggests that, despite RHPS4 treatment, cells pass through S-phase normally without any significant delay (Figure 1c,d).

RHSP4 Induces Replicative Stress and DNA Damage at Genomic and Telomeric Level in ALT Cells
Despite the absence of any significant S-phase progression delay, we found that RHPS4 treatment significantly increased replicative stress at both genomic and telomeric levels in ALT cells. Immunofluorescence staining of S33-phospho-RPA2 revealed a significantly higher number of foci in ALT cells treated than in untreated controls (Figure 2a,b). Similarly, replication stress at telomeres, evaluated by the analysis of telomere and S33-phospho-RPA2 signals, was also significantly increased by RHPS4 in ALT cells, whereas any response was observed in HOS cells (Figure 2a,c). Levels of replication stress at telomeres were further analyzed by the scoring of multi-telomeric signals (telomere doublets) after CO-FISH staining. In particular, double color stained doublets were evaluated to score the frequency of replication stress leading to recombinational events at telomeres. As also observed for telomere and S33-phospho-RPA2 colocalizations, we found that telomeric doublet frequency was also higher in U2OS and SAOS-2 than in HOS cells, confirming the evidence of higher basal replication stress at telomeres in ALT cells. In addition, RHPS4 treatment was able to increase very significantly (p < 0.01) the frequency of telomeric doublets in ALT cells (2 and 1.5-fold increase in U2OS and SAOS-2, respectively; Figure 2d,e) further confirming the higher sensitivity of ALT cells to G4 ligand-induced replication stress in telomeric regions. To evaluate if replication stress induction was also coupled by an increase in DNA damage at telomeres, the presence of co-localization of the telomere-associated factor TRF1 and the DNA damage marker 53BP1 was evaluated. Unexpectedly, no increase of DNA damage was detected in HOS cells after treatment with the compound neither at genomic nor telomeric level. Conversely, ALT-positive cell lines displayed a significant increase in DNA damage mostly in colocalization with telomeric clusters (Figure 3a,b). In particular, in U2OS cells, the mean of clusterized dysfunctional telomeres increased from 0.5 to 0.8, while in SAOS-2 cells, it increased from 2.6 to 3.8 (Figure 3a

RHPS4 Enhances APBs, T-SCE, and C-circles in ALT Cells
In order to assess whether or not RHPS4 was able to affect the ALT mechanism, the cardinal ALT hallmarks were analyzed. Surprisingly, APBs analysis showed that HOS cells did not display at all PML nuclear bodies. On the contrary, U2OS and SAOS-2 showed a massive presence of this protein at telomeres, as demonstrated by the co-localization of telomeres with PML. Following treatment, the percentage of telomeres colocalizing with PML foci increases in both U2OS and SAOS-2 (1.3-fold increase for both), but not in HOS (Figure 4a,b). In agreement with previously published data [47,50,51], we found that U2OS and SAOS-2 exhibited high levels of TERRA in contrast to telomerase positive HOS cells. However, RHPS4 was not able to alter TERRA levels, as evaluated by RNA FISH and qPCR analysis for subtelomeric regions of chromosomes 10, X, and Y (Figure 4c,d). The strongest evidence supporting the effect of RHPS4 on the ALT pathway comes from the analysis of T-SCE and c-circles. Indeed, RHPS4 determined a significant increase of T-SCE frequency in both ALT cell lines. In particular, U2OS and SAOS-2 showed a 7.3 and a 6.5-fold increase, respectively, whereas no modulation was observed in HOS (Figure 5a,b). Increased recombination rates were not observed at genomic level, as shown by G-SCE analysis (Figure 5a,c). In accordance with T-SCE, a 1.6-fold increase in c-circles amount was found in both U2OS and SAOS-2 upon RHPS4 treatment, whereas the HOS cell line did not show any modulation (Figure 5d,e).

RHPS4 Enhances APBs, T-SCE, and C-circles in ALT Cells
In order to assess whether or not RHPS4 was able to affect the ALT mechanism, the cardinal ALT hallmarks were analyzed. Surprisingly, APBs analysis showed that HOS cells did not display at all PML nuclear bodies. On the contrary, U2OS and SAOS-2 showed a massive presence of this protein at telomeres, as demonstrated by the co-localization of telomeres with PML. Following treatment, the percentage of telomeres colocalizing with PML foci increases in both U2OS and SAOS-2 (1.3-fold increase for both), but not in HOS (Figure 4a,b). In agreement with previously published data [47,50,51], we found that U2OS and SAOS-2 exhibited high levels of TERRA in contrast to telomerase positive HOS cells. However, RHPS4 was not able to alter TERRA levels, as evaluated by RNA FISH and qPCR analysis for subtelomeric regions of chromosomes 10, X, and Y (Figure 4c,d). The strongest evidence supporting the effect of RHPS4 on the ALT pathway comes from the analysis of T-SCE and c-circles. Indeed, RHPS4 determined a significant increase of T-SCE frequency in both ALT cell lines. In particular, U2OS and SAOS-2 showed a 7.3 and a 6.5-fold increase, respectively, whereas no modulation was observed in HOS (Figure 5a,b). Increased recombination rates were not observed at genomic level, as shown by G-SCE analysis (Figure 5a,c). In accordance with T-SCE, a 1.6-fold increase in c-circles amount was found in both U2OS and SAOS-2 upon RHPS4 treatment, whereas the HOS cell line did not show any modulation (Figure 5d,e).

RHPS4 Decreases the Levels of RAD51 and CHK1 Proteins
Since CHK1 and RAD51 are well-known players in homologous recombination and were also downregulated upon RHPS4 treatment, as previously reported in glioma cells [33]; we decided to evaluate their levels, in order to gain insight in the recombinational mechanism activated. We noticed a significant 34% and 32% of RAD51 protein decrease in U2OS and SAOS-2, respectively ( Figure  6a,b). Similar reduction was also observed for CHK1 (Figure 6a,b). As reported in Figure 1c

RHPS4 Decreases the Levels of RAD51 and CHK1 Proteins
Since CHK1 and RAD51 are well-known players in homologous recombination and were also downregulated upon RHPS4 treatment, as previously reported in glioma cells [33]; we decided to evaluate their levels, in order to gain insight in the recombinational mechanism activated. We noticed a significant 34% and 32% of RAD51 protein decrease in U2OS and SAOS-2, respectively (Figure 6a,b). Similar reduction was also observed for CHK1 (Figure 6a,b). As reported in Figure 1c,d, BrdU incorporating cells quantified through cytofluorimetric analysis exclude that the RAD51 and CHK1 protein reduction upon RHPS4 treatment was due to a depletion of the S-phase. incorporating cells quantified through cytofluorimetric analysis exclude that the RAD51 and CHK1 protein reduction upon RHPS4 treatment was due to a depletion of the S-phase.

Discussion
In the last 20 years, telomeric G4 ligands have been proposed as telomere targeting agents able to rapidly induce telomere dysfunction and growth inhibition in a number of cancer cells both in vitro and in vivo. Interestingly, different G4 ligands (such as quinoline based-ligands, RHPS4, TMPyP4, pyridostatin, BRACO-19, and telomestatin) have been proven to be effective not only in telomerase positive but also in ALT-positive cancer cells [23][24][25][26]52,53]. To find a rationale supporting the observed cell growth inhibitory effect, some authors have raised the possibility that G4 stabilization in telomeric regions might inhibit the ALT-mediated recombination mechanism [35,36,54,55]. Conversely, more recently, other authors reported that G4 stabilizers are able to fuel the ALT mechanisms (both in a RAD51 dependent or independent manner) through the induction of replicative stress and DNA damage at telomeres, in particular in cells harboring ATRX mutations such as ALT cells [37,38]. In the present work, the effect of RHPS4, a well-known and potent telomeric G4 stabilizer, was evaluated in U2OS, SAOS-2 (ALT-positive), and HOS (telomerase positive/ALTnegative) osteosarcoma cell lines, in terms of cell growth inhibition, cell cycle progression, and modulation of the cardinal ALT hallmarks. In agreement with results obtained in ALT positive GM847DM cells [23], RHPS4 was able to reduce cell growth also in U2OS and SAOS-2 osteosarcoma cells (IC50 values: 1.4 and 1.6 μM, respectively). Interestingly, G4 stabilization has been recently proposed as a strategy for the selective targeting of ATRX-deficient gliomas [56]. Indeed, the ATRX protein has been implicated in the direct resolution of G4 secondary structures through its helicase Snf2 domain [57,58] and in the inhibition of RNA-DNA hybrids (R-loops) during transcription that favor G4 formation in the untranscribed DNA strand [59]. In osteosarcoma cells, RHPS4 effectiveness seems to be unlinked from both the genetic status of ATRX and the active TMM, as demonstrated by the very similar sensitivity of HOS telomerase-positive cells to the compound (IC50 value: 1.2 μM). Despite the similar sensitivity, we found that RHPS4 was able to induce genomic and, in particular, telomeric replicative stress preferentially in ALT-positive cell lines. ALT cancer cells exhibit higher basal levels of replication defects, leading to persistent DNA damage at telomeres [37,38]. Abrogation of the G2/M checkpoint, frequently observed in these tumors, may allow cell cycle progression in the presence of incomplete DNA replication, triggering telomere synthesis in the M-phase through the BIR pathway [38]. In particular, we found that S33-p-RPA2/telomere colocalization and telomeric doublets events were significantly increased, indicating that replicative stress in an ALT-permissive genetic background results in telomeric anomalies and possibly DNA damage. Consistent with this observation, analysis of TIFs and APBs confirmed that RHPS4 was able to induce DNA damage in ALT cells, increasing the frequency of DNA damage in clusterized telomeres and PML/telomere colocalizations. Ultrabright telomeric signals are a well-known marker of ALT and it is widely accepted that, in ALT cells, damaged telomeres converge in nuclear subcompartments together with

Discussion
In the last 20 years, telomeric G4 ligands have been proposed as telomere targeting agents able to rapidly induce telomere dysfunction and growth inhibition in a number of cancer cells both in vitro and in vivo. Interestingly, different G4 ligands (such as quinoline based-ligands, RHPS4, TMPyP4, pyridostatin, BRACO-19, and telomestatin) have been proven to be effective not only in telomerase positive but also in ALT-positive cancer cells [23][24][25][26]52,53]. To find a rationale supporting the observed cell growth inhibitory effect, some authors have raised the possibility that G4 stabilization in telomeric regions might inhibit the ALT-mediated recombination mechanism [35,36,54,55]. Conversely, more recently, other authors reported that G4 stabilizers are able to fuel the ALT mechanisms (both in a RAD51 dependent or independent manner) through the induction of replicative stress and DNA damage at telomeres, in particular in cells harboring ATRX mutations such as ALT cells [37,38]. In the present work, the effect of RHPS4, a well-known and potent telomeric G4 stabilizer, was evaluated in U2OS, SAOS-2 (ALT-positive), and HOS (telomerase positive/ALT-negative) osteosarcoma cell lines, in terms of cell growth inhibition, cell cycle progression, and modulation of the cardinal ALT hallmarks. In agreement with results obtained in ALT positive GM847DM cells [23], RHPS4 was able to reduce cell growth also in U2OS and SAOS-2 osteosarcoma cells (IC50 values: 1.4 and 1.6 µM, respectively). Interestingly, G4 stabilization has been recently proposed as a strategy for the selective targeting of ATRX-deficient gliomas [56]. Indeed, the ATRX protein has been implicated in the direct resolution of G4 secondary structures through its helicase Snf2 domain [57,58] and in the inhibition of RNA-DNA hybrids (R-loops) during transcription that favor G4 formation in the untranscribed DNA strand [59]. In osteosarcoma cells, RHPS4 effectiveness seems to be unlinked from both the genetic status of ATRX and the active TMM, as demonstrated by the very similar sensitivity of HOS telomerase-positive cells to the compound (IC50 value: 1.2 µM). Despite the similar sensitivity, we found that RHPS4 was able to induce genomic and, in particular, telomeric replicative stress preferentially in ALT-positive cell lines. ALT cancer cells exhibit higher basal levels of replication defects, leading to persistent DNA damage at telomeres [37,38]. Abrogation of the G2/M checkpoint, frequently observed in these tumors, may allow cell cycle progression in the presence of incomplete DNA replication, triggering telomere synthesis in the M-phase through the BIR pathway [38]. In particular, we found that S33-p-RPA2/telomere colocalization and telomeric doublets events were significantly increased, indicating that replicative stress in an ALT-permissive genetic background results in telomeric anomalies and possibly DNA damage. Consistent with this observation, analysis of TIFs and APBs confirmed that RHPS4 was able to induce DNA damage in ALT cells, increasing the frequency of DNA damage in clusterized telomeres and PML/telomere colocalizations. Ultrabright telomeric signals are a well-known marker of ALT and it is widely accepted that, in ALT cells, damaged telomeres converge in nuclear subcompartments together with PML protein and other proteins that mediate homologous recombination (HR) and/or break-induced repair (BIR) [60]. The increase of telomeric DNA damage and APBs was not coupled by an increase of TERRA expression, as shown by RNA-FISH and qPCR experiments, but was strongly supported by T-SCEs analysis. Data showed that T-SCEs (but not G-SCEs) greatly increased after RHPS4 treatment in both U2OS and SAOS-2 cells, whereas no changes were observed in HOS cells as well as in telomerase positive glioblastoma U251MG cells (data not shown). To further corroborate these data c-circles analysis, we also considered the most reliable ALT marker for diagnostic purposes [61], which were significantly induced in ALT-positive cells but not in the HOS cell line. Taken together, ALT hallmark analysis indicates that telomeric replicative stress induced by the G4 ligand RHPS4 is able to increase DNA damage at telomeres fueling the ALT pathway in ALT-positive osteosarcoma cells. However, our data do not clarify which kind of ALT pathway is activated in response to RHPS4. Very recently, we showed that RHPS4 is able to strongly downregulate CHK1 and RAD51 proteins in glioblastoma cancer cells, probably due to specific targeting of G4 located in promoters of these genes [33]. Interestingly, RAD51 is implicated in HR-mediated telomere lengthening in ALT cells but not in BIR [62]. Protein level analysis showed a 30-35% reduction of RAD51 in both ALT-positive cell lines, suggesting that HR may be partially inhibited and pointing to a major role of BIR in response to RHPS4. Recently the group of Jerry Shay demonstrated that pyridostatin treatment is able to induce BIR in ALT cells, and our data seem to go in the same direction [38]. However, which kind of ALT pathway is activated in response to RHPS4 is still unclear to us, and further experiments are needed to clarify this aspect.

Conclusions
In conclusion, the present work indicates that, despite its efficacy to inhibit cell growth, the G4-ligand RHPS4 is able to stimulate telomeric recombination and c-circles formation through the induction of telomeric replicative stress, telomeric DNA damage, and APBs formation in U2OS and SAOS-2 ALT osteosarcoma cells. These are in accordance with with the emerging role of G4 secondary structures and telomeric replicative stress as triggers of the recombination-mediated telomere lengthening in an ALT permissive (ATRX-deficient) genetic background [37,38]. ALT fueling by G4 stabilizing compounds raises questions on their possible employement as therapeutic options to target ALT-positive tumors, as recently proposed for gliomas [56].